Neuron Explorer

Qwen3-0.6B

'and' closing a list of items

A neuron in layer 15 of Qwen3-0.6B. Its strongest snippets are very consistent: they mostly peak on the same token. It is featured, so it has steering outputs.

What this neuron does

This neuron fires on the word 'and' when it joins the last two items in a list, such as 'Photography, Film, and Video' or 'sand, loam and clay'. Its negative side reacts to commas that start a new clause. Turning it up made the model drift into long, comma-heavy sentences.

Written by Claude from this neuron's activations and steering outputs. Confidence: medium. Think it is wrong? Flag it (opens a prefilled issue on GitHub).

Fires most on

The text windows where this neuron's activation was highest. Amber marks positive activation; the outlined token is the peak. Hover a token for its strength.

  1. #1 peak

    ography Visual Data: Photography, Film, and Video Using Documents as Data Qualitative Online Research: Using the Internet Observation and Med

  2. #2 peak

    is similar to use of arrowleaf balsamroot (pronghorn, bighorn sheep, and mule deer). Arrowleaf balsamroot is grazed year-round but

  3. #3 peak

    be flattered (if not shocked) that you'd be insightful enough to ask.⏎By Rod Machado<|endoftext|>PARIS — Should Palmyra be rebuilt? And under what conditions? No sooner had Syrian forces and the Russian Army freed the "

  4. #4 peak

    court in France. The book describes his court and the close relationships between the British and French royal families.⏎King in Exile: James II: Warrior, King and Saint by John Callow. Reassesses James's strategy for dealing with his

  5. #5 peak

    operation conditions, and inputs and outputs within and between each unit, vary in practice. These process variations can impact the key performance outcomes (titer, rate, and yield), which determine economic viability when the process is scaled up. The following descriptions

  6. #6 peak

    performed in the Globe Theatre. However, it was not only a period of prosperity, as there were a lot of religious conflicts (Catholics, Protestants and Puritans) with the resulting emigration to America of many Puritans

Steering

Push this neuron up or down while the model writes, and watch what changes. The outputs were generated ahead of time, so nothing runs in your browser.

turned down a lot
Prompt

Here is something I have been thinking about lately: What if, if you don't do something, do not do something else. What if, if you don not do something. If you not do something, do not do something. If you not do something, do not do something,

Most negative on

Where the activation went furthest below zero. Blue marks negative activation. Many neurons mean something different in each direction.

  1. #1 peak

    exceed two feet. It’s rust-colored, with a long, slightly downwardly curved bill that’s two-toned — dark above, a dull orange below. The neck is rather long but thick. The back is

  2. #2 peak

    or CD-ROM. Includes the Catholic Encyclopedia, Church Fathers, Summa, Bible and more all for only $19.99...⏎A Spanish historian; born at Cuellar, in the

  3. #3 peak

    = 871) from 5 schools attended one of 2 DST performances. Participants demonstrated increased knowledge on 5 fact-based questions (mean improvement range, 19%- 35%; p < .001 for all

  4. #4 peak

    night to make it to Willamette’s 1929 freshman orientation. A month later the stock market collapsed — his bank failed, his savings account was wiped out and his scholarship was withdrawn for lack of funds.

Connections

Show on map

Which neurons in earlier layers feed this one, and which later ones it feeds. Amber links excite, blue links suppress; longer bars are stronger.

Fed by earlier layers

  1. L13:1632 suppresses words inside headlines and titles
  2. L13:1180 suppresses
  3. L12:2605 excites
  4. L14:1628 suppresses
  5. L13:1026 suppresses
  6. L11:559 excites
  7. L5:2313 excites
  8. L10:1459 suppresses
  9. L8:1473 suppresses
  10. L10:371 excites
  11. L11:45 suppresses
  12. L14:2657 excites
  13. L11:840 weakly excites
  14. L9:2357 weakly excites
  15. L14:2899 weakly suppresses
  16. L13:1610 weakly excites
  17. L10:1794 weakly excites
  18. L12:3061 weakly suppresses
  19. L11:514 weakly excites
  20. L11:237 weakly excites
  21. L12:1959 weakly suppresses
  22. L10:1643 weakly suppresses
  23. L6:676 weakly suppresses
  24. L11:1968 weakly suppresses

Feeds later layers

  1. L16:2577 strongly excites
  2. L17:2221 strongly excites
  3. L16:1145 excites commas inside short lists
  4. L20:427 suppresses
  5. L16:650 excites
  6. L20:373 suppresses
  7. L16:1463 suppresses
  8. L21:739 excites 'both X and Y' pairing words
  9. L18:1327 suppresses
  10. L16:2432 excites
  11. L19:1414 suppresses
  12. L26:347 excites the 're' in self-reliant and similar
  13. L24:205 excites
  14. L17:2234 suppresses
  15. L19:1245 excites comma opening a list of parallel items
  16. L19:1262 excites
  17. L26:236 suppresses
  18. L18:1498 suppresses
  19. L26:65 suppresses
  20. L19:2451 suppresses
  21. L18:862 suppresses hedges like 'it turns out' and 'if any'
  22. L16:1669 suppresses
  23. L17:572 suppresses
  24. L17:2610 excites and or comma closing a list

Activation histogram

How often this neuron fires at each strength. Almost always it does close to nothing (the tall bars in the middle); the snippets above are the rare moments it fired hard, amber for positive and blue for negative.

1 values between -5.15 and -4.981 values between -4.98 and -4.811 values between -4.81 and -4.641 values between -4.47 and -4.301 values between -3.96 and -3.791 values between -3.45 and -3.281 values between -3.28 and -3.111 values between -3.11 and -2.942 values between -2.94 and -2.771 values between -2.77 and -2.603 values between -2.60 and -2.433 values between -2.43 and -2.272 values between -2.27 and -2.107 values between -2.10 and -1.938 values between -1.93 and -1.765 values between -1.76 and -1.599 values between -1.59 and -1.429 values between -1.42 and -1.258 values between -1.25 and -1.0816 values between -1.08 and -0.9123 values between -0.91 and -0.7422 values between -0.74 and -0.5732 values between -0.57 and -0.4086 values between -0.40 and -0.23221 values between -0.23 and -0.061008 values between -0.06 and 0.111449 values between 0.11 and 0.28743 values between 0.28 and 0.45193 values between 0.45 and 0.6237 values between 0.62 and 0.794 values between 0.79 and 0.962 values between 0.96 and 1.135 values between 1.13 and 1.302 values between 1.47 and 1.64offnegativepositiveoften

Logit lens

The words this neuron nudges the model toward (amber) or away from (blue) when it fires. A hint, not a label: many neurons work through later layers instead.

Promotes

  1. and
  2. 和完善
  3. and
  4. anden
  5. 时间和
  6. 性和
  7. enschaft
  8. && ⏎
  9. ()'
  10. 清代

Suppresses

  1. ;
  2. ,“
  3. ;
  4. 其它
  5. ;}⏎⏎
  6. ;]/
  7. ,
  8. (Connection
  9. emies
  10. ٬